Instagram spent 2026 turning its comment section into something you can no longer read as a column of strings. Photo comments — replying to a post or Reel with an image from your camera roll — rolled out in phases from mid-2026, and they change what a comment section is for. If you analyse comments, run giveaways, or handle UGC for a brand, this guide covers what a photo comment looks like when you download Instagram comments free to CSV, Excel or JSON, what does and does not come with it, and how to work with a thread whose punchline is a picture.
Instagram's photo comments, in one paragraph
Open the comment field on a post or Reel and, if the feature has reached your account, there is a photo icon beside it. Tap it, choose an image from your camera roll, optionally type a caption, and the comment posts with the picture embedded in the thread. It works on ordinary posts and on Reels. The rollout has been phased since around July 2026, which is why the feature appears and disappears between accounts and why half the internet believes Instagram removed it. Notably, when Instagram added comment editing earlier in 2026, editing was limited to the text portion — the image in a photo comment cannot be changed after posting.
Why an image reply breaks the way you read a comment section
The reason this matters is not novelty. It is that the highest-signal comments on a post are now the ones a spreadsheet cannot see.
Think about the effort curve. Typing "love this" costs two seconds. Posting a photo means leaving the keyboard, opening the camera roll, finding the right image, and posting it — call it fifteen to thirty seconds and a deliberate decision. People do not do that casually. In practice the photo replies under a post are the receipt someone posted to prove they bought the thing, the screenshot of the thing going wrong, the before-and-after, the meme that lands the joke. They are the comments you would most want in a report, and they are the comments that arrive with an empty text field.
So the naive analysis fails in a specific direction. A sentiment pipeline scores the blanks neutral and reports a flatter, calmer comment section than the real one. A keyword count for your product name misses every person who answered by photographing the product. A word cloud drops the most committed responses on the post. None of this is loud enough to notice — the numbers still look plausible, they are just wrong in a consistent direction.
Downloading the comments on a post that has photo replies
Mechanically nothing changes. One setting matters more than it used to.
- Copy the URL. Three dots, Copy link, or the address bar. A
/p/link, a/reel/link and a share link with tracking junk all resolve to the same media. - Paste it into the Instagram comment exporter. It reads the media's comment list, not the fifteen comments the page renders before you tap "View more comments". No login, no extension.
- Turn replies on. This is the one that matters. Photo comments are overwhelmingly replies — they answer a question, they respond to a claim, they escalate a joke. Replies are off by default because expanding them multiplies your row count, so a top-level-only download of a lively post will miss most of the visual conversation entirely. The bulk export guide covers what that does to volumes.
- Choose a format. Excel to filter by hand, JSON to hand to a script, CSV for everything else.
- Sort by text ascending. Every caption-free image reply collects at the top of the sheet in one click.
Reading around the gaps: three columns that recover the meaning
The image is not in the file. You can still get most of the way to the meaning without opening the app three hundred times, using columns that are in the file.
likes is your triage. A caption-free photo reply with 340 likes is doing something significant on that post. One with zero is somebody's cat. Sort the blank-text rows by likes descending and the top ten are almost always worth opening manually. This single move is the difference between "we cannot analyse image comments" and "we looked at the ten that mattered".
reply_to_id gives you the context the picture is answering. Every reply row carries the id of its parent comment, so you can look up the text of the comment being answered even when the answer itself is an image. "Does this actually fit a size 14?" followed by a picture is a comment you can interpret without ever seeing the photo. Half the meaning of an image reply lives in the comment above it, and that comment is text.
created_at tells you what triggered the wave. Image replies clump. A cluster of them fifteen minutes after a creator posted a story pointing at the post, or two hours after a shipment landed, is a pattern you can see in a timestamp column and cannot see by scrolling.
Beyond those, every download carries author, username, avatar_url, text, replies, language, is_pinned and id, plus Instagram's own raw fields such as verification status and whether the post's owner liked the comment. For sorting out which handles to actually talk to, Instagram comment sentiment analysis covers the text side of the same file.
Photo comments and giveaway rules
Giveaway hosts get a version of this problem with a deadline attached.
If your rules say "comment the word BLUE to enter" and somebody posts a photo of something blue, they believe they have entered. Their row has no text, no keyword filter will match it, and they will be in your replies within the hour asking why they were excluded. The complaint is reasonable — from where they sit, they answered.
Decide before the giveaway opens, not after:
- Require typed entries. One sentence in the official rules, and every entry becomes machine-checkable. The right call for anything over a few hundred entries.
- Allow image entries and review them by hand. Download with replies on, filter to blank-text rows, and check them individually. Fine at twenty, unpleasant at four hundred.
Either way, publish the eligible entry count next to the winner — a draw is only verifiable if a stranger knows how many entries were in the pool. The Instagram giveaway entries guide covers deduplicating by handle and enforcing a closing timestamp, and the US FTC endorsement guidance covers disclosure once entrants are posting on your behalf.
Brands, UGC and the accidental proof-of-purchase archive
There is an upside here for anyone doing brand work, and it is worth naming because it is the reason to care rather than merely cope.
Photo comments have quietly turned comment sections into unsolicited UGC pipelines. People are posting pictures of your product on your own post, unprompted, with no campaign brief and no incentive. That is the material brands normally pay agencies to source. The comment list gives you the handles and the timestamps of everyone who did it — which is exactly what you need to go and ask permission properly.
One thing to be firm about: a picture posted in a public comment is not a licence to use it in an ad. Reposting UGC without explicit permission is a routine and expensive mistake, and it is a copyright question, not a platform-policy question — the photographer owns the photo. The workflow that works is the boring one: export the comment list, identify the handles that posted images, ask each of them in writing, keep the reply. On the data side, if you are storing those handles and images in a system, the GDPR framework and California's CCPA both apply to identifiable people, whatever platform the data came from. Instagram's own Help Centre is the source for the comment controls available to you as the post owner.
Where this guide comes from
We build and run the exporter, so what we can speak to is what we observe in aggregate from operating it, not any individual account's numbers.
The blank-text row is not a new phenomenon to us, and it is why our Instagram export was built to keep every raw field the platform returns alongside the friendly columns rather than reducing a comment to its text. Comments have been arriving without text for as long as people have been posting stickers on other platforms; Instagram photo comments simply made the category big enough that ordinary users notice and write in about it.
The most common support question we get on Instagram, by a wide margin, is why the row count is lower than the comment number displayed on the post. It is worth being precise: the badge counts comments the platform has accepted over the post's life, while a download returns comments the comment list currently serves. Comments from accounts that have since been deactivated, restricted or made private are counted in the first number and absent from the second. It is not a truncated file. Notably this gap runs in the opposite direction on YouTube for different reasons, which is why we do not publish a single rule of thumb for it.
What we will not claim is that we can hand you the image. We cannot, and any tool that says it can is describing something else. What the file honestly gives you is the identity, the timing, the engagement, the thread position and the parent comment for every image reply on the post — enough to find the twelve that matter in a post of three thousand.
The column names above, the free cap of 100 comments per post, and the replies-off-by-default behaviour are the live product values at the time of writing.
Download Instagram comments free →
Related reading: Instagram comment downloader, how to analyze Instagram comments, export Instagram comments to CSV.